Enhanced Conformational Sampling Method Based on Anomaly Detection Parallel Cascade Selection Molecular Dynamics:...

Area of Science:

• Computational Biology
• Biophysics
• Molecular Dynamics

Background:

• Generating protein conformational transition pathways is crucial for understanding biological functions.
• Conventional methods struggle with rare events and require extensive simulation times.
• Predicting product states a priori limits the scope of pathway discovery.

Purpose of the Study:

• To develop a novel rare-event sampling method, ad-PaCS-MD, for efficient exploration of conformational transition pathways.
• To enable pathway searching without prior knowledge of the product state.
• To enhance conformational sampling efficiency compared to conventional methods.

Main Methods:

• Proposed anomaly detection parallel cascade selection molecular dynamics (ad-PaCS-MD).
• Integrated anomaly detection generative adversarial network (anoGAN) to identify and resample rare, essential protein configurations.
• Utilized Markov state model construction for free energy landscape analysis.

Main Results:

• ad-PaCS-MD successfully promoted open-closed transitions in T4 lysozyme and maltose-binding protein.
• Large-amplitude transitions were achieved within nanosecond-order simulation times.
• Demonstrated significantly higher conformational sampling efficiency than conventional molecular dynamics (CMD).

Conclusions:

• ad-PaCS-MD offers an efficient approach for discovering protein conformational transition pathways, particularly for rare events.
• The method automates the promotion of rare events by focusing on anomalous states.
• ad-PaCS-MD provides a powerful tool for exploring protein dynamics and free energy landscapes.